Content uploaded by Dipanjan Nag
Author content
All content in this area was uploaded by Dipanjan Nag on Dec 19, 2019
Content may be subject to copyright.
Full Terms & Conditions of access and use can be found at
https://www.tandfonline.com/action/journalInformation?journalCode=ttrv20
Transport Reviews
ISSN: 0144-1647 (Print) 1464-5327 (Online) Journal homepage: https://www.tandfonline.com/loi/ttrv20
Assessing urban sidewalk networks based on
three constructs: a synthesis of pedestrian level of
service literature
Dipanjan Nag, Arkopal Kishore Goswami, Ankit Gupta & Joy Sen
To cite this article: Dipanjan Nag, Arkopal Kishore Goswami, Ankit Gupta & Joy Sen (2019):
Assessing urban sidewalk networks based on three constructs: a synthesis of pedestrian level of
service literature, Transport Reviews, DOI: 10.1080/01441647.2019.1703841
To link to this article: https://doi.org/10.1080/01441647.2019.1703841
View supplementary material
Published online: 18 Dec 2019.
Submit your article to this journal
View related articles
View Crossmark data
Assessing urban sidewalk networks based on three constructs:
a synthesis of pedestrian level of service literature
Dipanjan Nag
a
, Arkopal Kishore Goswami
a
, Ankit Gupta
b
and Joy Sen
a,c
a
Ranbir & Chitra Gupta School of Infrastructure Design & Management, Indian Institute of Technology
Kharagpur, Kharagpur, India;
b
Department of Civil Engineering, Indian Institute of Technology, Banaras Hindu
University, Varanasi, India;
c
Department of Architecture & Regional Planning, Indian Institute of Technology
Kharagpur, Kharagpur, India
ABSTRACT
Pedestrian Level of Service (PLOS) models are widely used to assess
walking facilities. These models have been in existence since the
1970s, wherein the process broadly consists of three steps, i.e.
attribute selection, model calibration, and classification of model
results into service-level categories, based on Measures of
Effectiveness (MOEs). This paper reviews existing sidewalk PLOS
studies based on their association with the three constructs of
flow characteristics,built environment and users’perception, which
in combination represents the entire walking environment
spectrum, as has been indicated by existing researchers. Forty-
seven PLOS studies, along with eight review papers, written by
authors from the Americas, Europe, Asia and Australia, between
the years of 1971 and 2019, are analysed in this review. The
review finds that although 49% of the studies employed both
qualitative and quantitative data for their respective
methodologies, but none of them use all the three broad
constructs in a combined fashion. Also, in selecting the attributes
to be used for developing the PLOS, these studies have only
referred to previous literature available at that point in time, and
not employed any consistent and robust method in selecting
context-specific attributes. When it came to the preferred analysis
technique, 60% of the studies favoured the use of the regression
technique while calibrating their model, whereas 22% used a
points-based marking scheme. Finally, 89% of the studies
manually classifies the PLOS model results to respective service
levels (i.e. letter grades), as opposed to utilising a classification
algorithm. In addition, this review could identify only one paper
that describes a PLOS based on pedestrian route directness, which
is a measure of pedestrian network connectivity. In view of these
findings, the review paper suggests the need of a robust
methodology in selection of attributes and the use of innovative
modelling techniques, both of which could allow the utilisation of
all three constructs. Also, such advanced modelling techniques
could bypass the need for categorising service levels manually.
Finally, the study advocates the use of network connectivity
measures in developing sidewalk PLOS, as it is an important part
of the built environment.
ARTICLE HISTORY
Received 8 January 2019
Accepted 14 November 2019
KEYWORDS
Pedestrian Level of Service
(PLOS); flow characteristics;
built environment; users’
perception; sidewalk network
© 2019 Informa UK Limited, trading as Taylor & Francis Group
CONTACT Dipanjan Nag el.diablo.diablo78@gmail.com
TRANSPORT REVIEWS
https://doi.org/10.1080/01441647.2019.1703841
1. Introduction
Pedestrian Level of Service (PLOS) is defined by the Highway Capacity Manual (2000)as“a
qualitative measure of pedestrian trafficflow, along with environmental factors that might
affect perceived level of comfort, convenience, safety, security and the economy of
walkway systems”. PLOS allows six letter grades (A–F) to be assigned to pedestrian facili-
ties under consideration, where a score “A”would mean the best PLOS, and alternately the
score of “F”means the worst. This process of categorisation to different service levels
(letter grades) is based on certain attributes, also called measures of effectiveness
(MOE). Kadali and Vedagiri (2016) consider PLOS assessment as the most reliable
method for assessing existing facilities. While using this measure it is important to under-
stand the “criteria for assessment”as defined by different researchers. These “criteria for
assessment”focus on different aspects of the walking environment and can be grouped
under three broad constructs. Existing research and review studies on factors influencing
the entire range of the walking environment have hinted towards the following three
broad constructs (Cervero, Sarmiento, Jacoby, Gomez, & Neiman, 2009; Ewing &
Cervero, 2010; Landis, Ottenberg, Mcleod, & Guttenplan, 2001; Maghelal & Capp, 2011;
Olsen, Macdonald, & Ellaway, 2017; Singh & Jain, 2011; Vale, Saraiva, & Pereira, 2016).
.Flow characteristics of pedestrians and vehicular traffic
.Physical elements of built walking environment, and
.Users’perception of the walking environment
It is observed that there is a large variation in the methodology adopted in existing
PLOS studies. These differences are in terms of selection of attributes, the techniques
adopted for PLOS model calibration and finally in the classification to service levels. There-
fore, an in-depth review of existing literature is required on these aspects which is cur-
rently not being addressed by existing review papers. Thus, the present paper
systematically reviews a comprehensive list of PLOS studies with the timeline set from
1971 (i.e. the first PLOS conception by John J. Fruin) up till 2019 (i.e. the present), with
an intent to answer the following five questions:
.How are the attributes being selected while developing PLOS measures?
.How are these measures modelled or calibrated?
.How are the service levels defined?
.Are all the constructs represented?
.Is the walking network (network attributes) taken into account?
2. PLOS concept for evaluation of the urban walking environment: the
three constructs
PLOS could be considered equivalent to any performance assessment tool, where the
efficiency of the tool lies in its ability to capture all the aspect of the process. For
example, in the business or marketing fields (Franceschini, Cignetti, & Caldara, 1998)a
service quality measure, considers two groups of concepts –perception & expectations.
2D. NAG ET AL.
If an analogy were to be drawn between such measures and PLOS –perception would
relate users experience of the facility, whereas expectation would encompass all the oper-
ational and built characteristics of the facility.
Perception plays an important role in an individuals’decision to use a walking facility.
Research has shown that user perception factors such as absence of comfort, convenience,
safety, and shade (Zainol, Ahmad, Nordin, & Aripin, 2014), and physical built walking
environment elements such as the presence of driveways, bus stops, and number of
vehicle lanes (Choi, Kim, Min, Lee, & Kim, 2016), reduce the level of pedestrians’satisfac-
tion, which in turn, affects their decision to walk. Kim, Sohn, and Choo (2017) analysed
relationships between pedestrian traffic volume and various measures of the physical
built environment at the street and neighbourhood levels to show their effect on flow
characteristics.
Various review papers and reports have conveyed their concern regarding the ability of
existing studies to capture all the aspects of a facility evaluation process. Bloomberg and
Burden (2006) have explained about three components –sidewalk environment, ped-
estrian characteristics and flow characteristics. The sidewalk environment component is
related to the physical built walking environment, the pedestrian characteristics is associ-
ated to the individual socio-economic information, behaviour, physical ability and their
expectation (users’perception), and the flow characteristics is related to the pedestrian
and vehicular trafficflow characteristics. Figure 1 is closely adopted from Bloomberg
and Burden (2006) considering the three broad constructs and their interrelationships.
Singh and Jain (2011) have stated that assessment of facilities have not been inclusive
of the entire “walking environment”. Kadali and Vedagiri (2016) have advocated on the
Figure 1. Interrelationships between the three broad constructs.
TRANSPORT REVIEWS 3
creation of a combined methodology for quantitative (speed, density, etc.) and qualitative
(comfort, safety, etc.) assessments of the pedestrian facilities. Maghelal and Capp (2011)
have addressed the need for focusing on built environment attributes as they significantly
influence walking. As such, these three constructs are central to the evaluation procedure
of pedestrian facilities and future studies should consider integrating them.
3. PLOS and walkability indices
The Online Travel Demand Management (TDM) Encyclopaedia by Victoria Transport Policy
Institute (2009)defines “walkability”as overall walking conditions in an area and these
indices takes into account a large number of considerations based on the scale of evalu-
ation. Walkability indices are assessed at a specific site, a neighbourhood level or a com-
munity level. PLOS on the other hand always evaluates a segment of the walking facility
such as, sidewalks, mid-blocks or intersections.
In contrast to the three broad constructs explained above, Vale et al. (2016) identified
walkability indices as measures which include assessment of the infrastructure, topology,
proximity, and attraction potential of a facility. It appears that the purview of walkability
indices is much broader than PLOS measures and encompasses the three broad con-
structs. Walking as a mode is much more than just transportation between origin and des-
tination, it involves a social aspect as well. Since PLOS has emanated from vehicular LOS, it
has not been able to cater to this social aspect. An example for understanding this argu-
ment is the presence of vendors on sidewalk. Street vendors are the leading cause of side-
walk encroachment in developing countries that causes user discomfort, and also a
reduction in sidewalk capacity, when viewed from a traffic engineering standpoint.
However, from a social engineering understanding, removal of the vendors may affect
the social cohesion, interaction and therefore local economy, especially in the market
areas. Connected to this, is the observation by Jan Gehl, who suggested in his book
“Life between buildings: Using public spaces”(Campos, 2012), that built environment
designed down to the last detail, is an essential component in influencing the quality of
the walking environment. The deterioration of such quality may not decrease the
number of people walking (as it is intrinsic to human nature), but may reduce the social
cohesion and interaction.
Moreover, as presented later in this review (Table 2 and Appendix 1), none of the PLOS
studies utilised land use as an explanatory variable in their statistical modelling process.
The same has been seen to be true about connectivity of links in the pedestrian
network. Researchers (Hajrasouliha & Yin, 2015; Li & Tsukaguchi, 2005; Ozbil, Peponis, &
Stone, 2011) studying the impact of pedestrian movements due to landuse and connec-
tivity have statistically established that these attributes play a major role, which walkability
studies have always incorporated (Ewing & Cervero, 2001; Ewing, Hajrasouliha, Neckerman,
Purciel-Hill, & Greene, 2016; Park, Ewing, Sabouri, & Larsen, 2019). In addition, two review
studies on walkability indices (Maghelal & Capp, 2011; Vale et al., 2016) were examined
along with other PLOS studies, which addressed a total of 57 walkability indices. These
researchers pointed out that most walkability studies have included infrastructure
based assessment, commonly used in PLOS, as a part of their methodology.
Following these arguments, it could be suggested that PLOS studies measure fewer
aspects of the walking environment in comparison to walkability studies. However, we
4D. NAG ET AL.
see that PLOS are still being widely used in engineering assessments. Also, Vale et al.,
2016 have argued that the number of methodologies to measure walkability is as
varied as the number of researchers creating them. Hence, to compare the relationship
between vast number of walkability indices and PLOS measures is a topic for a different
review.
4. Methodology
The review identifies PLOS studies written in English, using the electronic databases (e-
databases) and research tools. Over 500 research items (journal article, conference pro-
ceedings, relevant reports, manuals and guidelines) including many duplicates were
initially obtained in the preliminary search. PLOS for different pedestrian facilities
were found, such as sidewalks, crosswalks, mid-block crossings, etc. Only peer-reviewed
journal papers, indexed conference proceedings and relevant technical documents (i.e.
grey literature) were considered. This paper focuses on reviewing literature on PLOS
assessment of only sidewalks (link level assessment), inclusive of the papers that docu-
ments “network-wide”sidewalk PLOS in the existing body of research. E-database
searches using the keyword “network-wide pedestrian level of service”returned only
one research paper (Stangl, 2012) related to the PLOS concept. It has been seen that
network-wide pedestrian facility assessments generally relates to connectivity indices
like Pedestrian Route Directness, Link-to-Node Ratio, etc. (Dill, 2004; Tal & Handy,
2012) and not to the PLOS concept, as was done by Stangl (2012). Finally, 46 research
items related to PLOS of sidewalks (links) and 1 PLOS study using network connectivity
measures were taken forward for full paper review and critical discussions. Studies were
also reviewed for their association with the three broad constructs and their primary
methodology for development.
Eight other review papers, six from the PLOS domain and two from the walkability
indices’arena were also included in this review. Thus in total, 55 studies were part of
this review. This paper builds upon the most current review paper by Raad and Burke
(2018), which considered 58 PLOS studies (including sidewalks and crosswalks). In this
review, 27 out of the 58 studies were included, as they focussed on sidewalks. An
additional 20 studies, which were not a part of the Raad and Burke (2018) review paper,
were included in this paper.
5. Findings from the review
The findings are grouped under five distinct sections –(i) overview of all 55 studies (47
PLOS studies and 8 review papers); (ii) attributes that are repeatedly used; (iii) broad con-
structs involved; (iv) analysis techniques and service level categorisation in the 47 PLOS
studies. In addition, an attempt is made to highlight the differences between studies orig-
inating from the developed versus developing economies.
5.1. Overview of studies
Four out of 47 (9%) PLOS studies are technical documents, 6 (13%) are conference pro-
ceedings from eminent conferences in India and around the world, and the remaining
TRANSPORT REVIEWS 5
37 (78%) are peer-reviewed journals articles in the CiteScore range of 0.1 to 3.67. Referring
to Figure 2, it is seen that majority of the work on this topic had been produced by India
(13 out of 47) and USA (12 out of 47).
This review includes recent studies (i.e. within the last two years, 2017–2019). Overall,
9 out of 47 studies (20%) are within this year range. Percentage of studies in the other
year ranges are –58% in 2010–2019, 24% in 2000–2009, 8% in 1990–1999, 8% in
1980–1989 and 2% in 1970–1979. Figure 3 suggests that there is a shift in PLOS
studies from the developed to the developing economies. The classification of econom-
iesisasperInternationalMonetaryFund(2018) standards. There is an increase in PLOS
Figure 2. Research items as per article type and country.
Figure 3. Distribution of PLOS studies across countries and decades.
6D. NAG ET AL.
studies in developing countries in the last decade (i.e. 2010–2019), which suggests a
rising awareness of walking as a sustainable mode of transport and the importance con-
ferred on evaluating walking environments for planning improvements. Such a trend is
only natural as the previous four decades has had a large number of studies from devel-
oped economies.
Twenty-three out of 47 studies (49%) used combined qualitative and quantitative
data for their respective methodology, in comparison to studies using only quantitative
(31%) or only qualitative (20%) data. In terms of data collection methods, most studies
used multiple techniques, including videography which was found to be widely used
for collecting flow parameters (24 studies), followed by response collection from individ-
uals (21 studies). The response collection process was further carried out in two ways –
response collection using questionnaire on site (17 out of 21) and response collection by
showing video clips of pedestrian facilities to the respondents (4 out of 21). Other tech-
niques used by the studies were visual assessments (7 studies) by the researchers and
physical environment audits (16 studies) that collects more detailed measurements of
features.
All eight review papers examined were articles from peer-reviewed journals in the Cite-
Score range of 0.2 to 2.05. Six out of eight review papers are specific to PLOS studies
whereas the remaining two are reviews related to walkability tools. Figure 4 shows
that India (3 out of 8) and USA (2 out of 8) has produced the most number of PLOS
review studies. Three out of the six reviews concluded that most existing studies do not
include disabled pedestrians in their consideration. Two out of the six studies converged
to the understanding that existing methods of PLOS were not capturing the entire spec-
trum of walking and therefore innovative approaches should be employed to capture the
complete walking experience.
The findings of the 47 PLOS studies and the 8 review papers are presented in Table 1,2
(excerpt of Appendix 1) and 3(excerpt of Appendix 2). For the complete details please
refer to Appendices 1 and 2.
Figure 4. Review papers as per Journal and country of author(s).
TRANSPORT REVIEWS 7
Table 1. Summary of the existing review papers studied.
Review Papers
Country of
author(s) Journal Topic
Year
range
Number of
studies Major conclusions
Sisiopiku, Byrd, and
Chittoor (2007)
USA Transportation Research
Record
Sidewalk PLOS 2000
to
2006
5.HCM (2000) often over estimates sidewalk LOS
.No methods consider quantitative and qualitative aspects in sufficient
details
Maghelal and Capp
(2011)
USA Journal of Urban and
Regional Information
Systems Association
Walkability indices 1994
to
2008
25 .Built environment variables are central in pedestrian studies
.Attributes of indices were classified into objective,subjective and
distinctive
Singh and Jain (2011) India Journal of Engineering
Research and Studies
Sidewalk PLOS 1996
to
2007
9.No methodology could be universally applied as they do not capture the
entire walking spectrum
.Methods should not be data driven
Asadi-Shekari,
Moeinaddini, and
Zaly Shah (2013)
Malaysia Transport Reviews Sidewalk PLOS 1971
to
2011
17 .Lack of studies considering disabled pedestrians
.Difficult to link evaluation methods to design processes
.Non-applicability to all hierarchy of streets
Gupta and Pundir
(2015)
India Transport Reviews Pedestrian flow
characteristics (as a
part of PLOS)
1971
to
2013
18 .Early PLOS studies used flow characteristics as attributes
.Disabled pedestrians not considered
.Flow characteristics varies with different context and culture
Kadali and Vedagiri
(2016)
India Transportation Research
Record
PLOS of sidewalks and
crosswalks
1971
to
2014
42 .Need to consider disabled pedestrians under mixed traffic conditions
.Need to consider Landuse in assessment process
.Need to use advanced modelling techniques
Vale et al. (2016) Portugal Journal of Transport and
Land Use
Walkability indices 1997
to
2013
32 .PLOS studies are classified under infrastructure based accessibility
measure
.Myriad of indices with varying methodology
Raad and Burke
(2018)
Australia Transportation Research
Record
PLOS of sidewalks and
crosswalks
1971
to
2016
58 .Disabled pedestrians not considered
.Need to consider Landuse in assessment process
.Connectivity is not considered
8D. NAG ET AL.
Table 2. Modelling information for PLOS studies reviewed (excerpt from Appendix 1).
Research items
Attribute
selection
technique
Modelling specifications
Analysis method
No. of locations/
sample size/
observations
Goodness of
fit/validation
measureDependent attribute Independent attributes
Broad construct/s involved: flow characteristics
Kim, Choi, Kim, and Tay
(2014)
CEL Evasive movements Effective width, volume, landuse characters MLR 28 locations; 468
video samples of 5
mins each
R
2
= 0.77 Validated with
perceived PLOS of 216
users
Sahani and Bhuyan
(2015)
CEL Flow rate, space, speed
and V/C ratio
Effective width of sidewalk HCM (2000) methodology
and Self Organising
Maps (SOM) using
Analytical Neural
Network (ANN)
NR Silhouette, Davies-Bouldin,
Clinski-Harabasz and
Dunn Index
Raghuwanshi and Tare
(2016)
CEL Average pedestrian
space
V/C ratio of pedestrians, vehicles, pedestrian
crossing time, and parking factors
MLR 9 street sections NR
Sahani and Bhuyan
(2017)
CEL Flow rate, space, speed
and V/C ratio
Effective width of sidewalk HCM (2000) methodology
and three clustering
algorithms –Affinity
Propagation, SOM in
ANN and Genetic
Algorithm (GA)
3764 pedestrians
observed
Clusters validated using a
number of index –
Silhouette, Davies-
Bouldin, Clinski-Harabasz
and Dunn
Cepolina, Menichini, and
Gonzalez Rojas (2018)
CEL Perceived comfort Interpersonal distances Local density method,
Voronoi diagram
395 pedestrians Validated in comparison
with HCM (2000)
Broad construct/s involved: built environment
Shekari and Zaly Shah
(2011)
Citing from 20
design
guidelines
from
developed
nations
NA Traffic speed, lanes, buffers, crossing
distance, and 13 other attributes
PBS Two collector urban
street
NA
Stangl (2012) CEL Route directness score Community block-size Pedestrian Route
Directness Index
8different block size
varying from
200 ft. X 200 ft. to
1000 ft. X 1000 ft.
NA
Asadi-
Shekari, Moeinaddini,
and Zaly Shah (2012)
CEL and design
guidelines
NA Model 1: Slope, elevator, curb ramp, and
7 other attributes
Model 2: Traffic speed, pavement,
markings and 17 other attributes
PBS 1 street in Singapore NA
(Continued)
TRANSPORT REVIEWS 9
Table 2. Continued.
Research items
Attribute
selection
technique
Modelling specifications
Analysis method
No. of locations/
sample size/
observations
Goodness of
fit/validation
measureDependent attribute Independent attributes
Asadi-Shekari et al.
(2014)
Citing from 20
design
guidelines
from
developed
nations
NA Traffic speed, lanes, buffers, mid block and 23
other attributes
PBS 1 street in the
campus –20
guidelines
reviewed
Not reported
Broad construct/s involved: users’perception
Sahani, Praveena, and
Bhuyan (2016)
CEL Overall satisfaction
score
Traffic, safety, comfort, maintenance and
aesthetics score
Multinomial Logit 1425 respondents Chi-square value =
505.5depicts good fit.
70% of data used for
model fit, remaining 30%
used to validate
Bivina, Parida, Advani,
and Parida (2018)
CEL NA Physical: Surface quality, width, obstruction,
vehicular conflict, continuity,
encroachment, crossing facility, security,
walk environment, comfort
PBS 2804 respondents
from 5 cities
Cronbach’s alpha > 0.7
assess internal
consistency
Bivina and Parida (2019) CEL Latent Exogenous:
Safety; Security;
Mobility &
infrastructure;
Comfort &
convenience
Latent Endogenous:
Perceived PLOS
scores from users
Traffic volume, traffic speed, police
patrolling, street lights, CCTV camera,
width of sidewalks, continuity,
encroachment, surface quality, pedestrian
amenities, bus shelter, cleanliness,
planning for disabled, obstruction
Structural Equation
Modelling
502 responses Normed Fit Index (NFI) =
0.92; Comparative Fit
Index (CFI) = 0.953;
Tucker Lewis Index (TLI)
= 0.939; Acceptable is
NFI, CFI, TLI > 0.9
Root Mean Square Error
(RMSEA) = 0.05;
Acceptable is RMSEA <
0.06
Broad constructs involved: users’perception + built environment
Christopoulou and
Pitsiava-Latinopoulou
(2012)
CEL NA Traffic factors, geometric/environmental
factors and pedestrian movement factors
PBS Application on one
location
Compared with 5 existing
methodology
qualitatively
Parvathi (2018) CEL Perceived user score Sidewalk condition, road characteristics,
interaction between pedestrians and other
modes, buffer, transit area and safety
PBS, inverse variance to
calculate weights
Over 100
respondents
NA
Zannat, Raja, and Adnan
(2019)
CEL Model 1: Perceived
PLOS scores from
users
Model 2: Perceived
PLOS scores from
users
Model 3: Perceived
roadway conditions
Model 1: Perceived roadway conditions –
accessibility, safety, comfort and
attractiveness
Model 2: Physical feature measurement –
Footpath (width of sidewalk, lighting, etc.)
carriageway (median width, guard rail etc.)
and transit facilities (bus bay, sign etc.)
Model 3: Physical feature measurement
Model 1: Ordered probit
Model 2: Marginal
effects
Model 3: MLR
Model 1: 413
responses
Model 2: 413
responses and
physical feature
survey from 88
points in the city
Model 3: 413
NR for any model
10 D. NAG ET AL.
responses and
physical feature
survey from 88
points in the city
Broad constructs involved: flow characteristics + built environment
Karatas and Tuydes-
Yaman (2018)
CEL Pedestrian volume,
assessment score
Density, walkway width, buffer area, shade,
enclosure, motor vehicle, maintenance,
conflicts
Minimum PLOS of the
three methodology or
weighted average
method
81 road segments NA
Broad construct/s involved: (a) flow characteristics (b) users’perception + built environment
Indian Highway Capacity
Manual (2018)
CEL Model 1: flow rate
Model 2: Speed
Model 3: Space
Model 4: Users’score
Model 1, 2, 3: Density
Model 4: QoS (Bivina et al., 2018): Physical
and users’characteristics
Model 1, 2, 3: LR
Model 4: Bivina et al.,
(2018)–PBS, weighted
average method
Model 1, 2, 3: sample
size = 951
Model 4: NR
NR
Broad construct/s involved: (a) users’perception (b) built environment + flow characteristics
Marisamynathan and
Lakshmi (2016)
CEL Overall satisfaction
level
Sidewalk surface, presence of guardrails and
barriers, traffic volume, sidewalk width
Step-wise MLR 540 respondents R
2
= 0.935; 90% data used
for model fit 10% for
validation. MAPE = 3.14%
and RMSE = 2.02%
Zhao et al. (2016) CEL Users’perception
scores
Flow rate, vehicular flow, effective width,
segregation facilities, frequency of barriers,
on-street parking, green looking ratio,
connected regions
Image processing and
Fuzzy neural network
87 sidewalks and
4300 responses
Accuracy = 0.94 compared
with the testing data set
Daniel et al. (2016) CEL Participants score Footpath width, road width, surface damage,
number of obstructions, pedestrian flow
traffic volume
MLR 391 respondents
from 25 roads
R
2
= 0.97, Validation –
average error 1.28%
Sahani et al. (2017) CEL Model 1: Overall
Satisfaction
Model 2: Perception
scores
Model 1: Platoon size, traffic, safety, comfort,
maintenance and aesthetic score
Model 2: width, pedestrian volume,
obstruction, motorised and non-motorised
volume
Factor analysis and
discriminant analysis for
selecting variables; step-
wise MLR for PLOS
Model; GP for LOS
classification
1825 respondents Model 1: NR
Model 2: R
2
= 0.972
Note: CEL: Citing from Existing Literature; LR: Linear Regression; MLR: Multiple Linear Regression; NA: Not Applicable; NR: Not Reported; PBS: Points-Based System
TRANSPORT REVIEWS 11
Table 3. PLOS service-level definitions and classification techniques from different studies (excerpt from Appendix 2).
Research items
Classification
technique used
MOE used for PLOS
classification
Service-level definitions
ABCDEF
Broad construct/s involved: flow characteristics
Polus et al. (1983) None Density (ped/m
2
) <0.60 0.61–0.75 C
1
0.75–1.25
C
2
1.26–2.00
Not defined Not defined Not defined
Area module (m
2
/ped) >1.67 1.66–1.33 C
1
1.33–0.80
C
2
0.80–0.50
Not defined Not defined Not defined
Average speed (m/min) 0–40 40–50 C
1
50–75
C
2
75–95
Not defined Not defined Not defined
Sahani and Bhuyan
(2013)
Affinity Propagation
Clustering
Volume (ped/s/m) ≤0.052 >0.052–0.065 >0.065–0.081 >0.081–0.095 >0.095–0.114 >0.114
Space (m
2
/ped) >17.78–14.42 >11.3–14.42 >8.24–11.3 >7.82–8.24 >5.3–7.82 ≤5.3
speed (m/s) >1.53 >1.36–1.53 >1.14–1.36 >0.93–1.14 >0.71–0.93 ≤0.71
V/C ratio ≤0.4 >0.4–0.57 >0.57–0.76 >0.76–0.9 >0.9–1.0 >1
Sahani and Bhuyan
(2017)
AP, GA-Fuzzy & SOM in
ANN Clustering
methods
Volume (ped/s/m) ≤0.061 >0.61–0.081 >0.081–0.104 >0.104–0.127 >0.127–0.146 >0.146
Space (m
2
/ped) ≥16.53 <16.53 to 13.06 <13.06–9.91 <9.91–7.25 <7.25–4.48 ≤4.48
Speed (m/s) >1.21 >1.03–1.21 >0.88–1.03 >0.78–0.88 >0.62–0.78 ≤0.62
V/C ratio ≤0.34 >0.34–0.52 >0.52–0.67 .0.67–0.84 >0.84–1.0 >1.0
Broad construct/s involved: built environment
Stangl (2012) None Pedestrian route
directness score
85–100% 45–84% 30–44% 23–29% 7–22% 0–6%
Asadi-Shekari et al.
(2012)
None PLOS score generated
from the Disabled
PLOS and General
PLOS
80–100 60–79 40–59 20–39 1–19 0
Broad construct/s involved: users’perception
Sahani et al. (2016) None Overall satisfaction score <1.5 1.5 to <2.5 2.5 to <3.5 3.5 to <4.5 4.5 to 5.5 >5.5
Bivina et al. (2018) None Model scores >125 100–125 75–99 50–74 25–49 <25
Bivina and Parida
(2019)
None Perceived PLOS score Not defined as per service levels
Broad constructs involved: users’perception + built environment
Christopoulou and
Pitsiava-
Latinopoulou
(2012)
None Assessment score 42–35 <35–28 <28–21 <21–14 <14–7<7–0
12 D. NAG ET AL.
Parvathi (2018) None Perception score 4.34502 to
5.64542
5.64543 to 6.49512 6.49513 to 7.79349 7.79350 to
9.09538
9.09539 to
10.39522
10.39523 to
11.8697
Zannat et al. (2019) None Perceived PLOS score Not defined as per service levels
Broad constructs involved: flow characteristics + built environment
Rastogi, Chandra,
and Mohan (2014)
None Space (m
2
/ped) >5.00 >2.22–5.00 >1.43–2.22 >1.00–1.43 >0.69–1.00 <0.69
Flow rate (ped/min/m) ≤18 >18–35 >35–51 >51–66 >66–73 >73
Karatas and Tuydes-
Yaman (2018)
None Volume (ped./15 min) Conceptual model proposed hence no service-level definition provided
Scores from visual
assessment
Broad construct/s involved: (a) flow characteristics (b) users’perception + built environment
Mori and
Tsukaguchi (1987)
None Volume (ped/min/m) <20 20–78 78–108 >108 Not defined Not defined
Density (ped/m
2
) <0.2 0.2–0.8 0.8–1.5 >1.5 Not defined Not defined
Overall evaluation Not defined as per service levels
Indian Road
Congress (2012)
None Volume (ped/min/m) ≤12 12–15 15–21 21–27 27–45 >45
Space (m
2
/ped.) >4.9 3.3–4.9 1.9–3.3 1.3–1.9 0.6–1.3 <0.6
Qualitative description –
users & built
Ideal walk
condition and
factors
affecting PLOS
minimal
Reasonable condition
exists, factors
affecting safety and
comfort exists
Basic condition but
significant factors
affecting safety and
comfort also exists
Poor condition,
safety and
comfort
minimal
Walking
condition
unsuitable
Severely
restricted
walking
environment
Indian Highway
Capacity Manual
(2018)
None (classified as per
landuse –only
commercial is shown
here)
Flow rate
(ped/min/m)
≤13 >13–19 >19–30 >30–47 >47–69 >69
QoS: Model score ≥124 <124–106 <106–70 <70–52 <52 Not defined
Broad construct/s involved: (a) users’perception (b) built environment + flow characteristics
Landis et al. (2001) None Respondents score ≤1.5 >1.5–2.5 >2.5–3.5 >3.5–4.5 >4.5–5.5 >5.5
Petritsch et al.
(2006)
None Respondents score ≤1.5 >1.5–2.5 >2.5–3.5 >3.5–4.5 >4.5–5.5 >5.5
Tan et al. (2007) None Participants’scores <2.0 2.0 to <2.5 2.5 to <3.0 3.0 to <3.5 3.5 to <4.0 4.0 and above
Dowling et al.
(2009)
None Participants’scores <1.5 1.5 to <2.5 2.5 to <3.5 3.5 to <4.5 4.5 to <5.5 5.5 and above
Daniel et al. (2016) None Respondents’
perception score
>8.5 >7.0–8.5 >6.0–7.0 >5.0–6.0 >4.0–5.0 ≤4.0
Sahani et al. (2017) Genetic Programming
(GP) clustering
Respondents’
perception score
≤1.8 >1.8 to 2.7 >2.7 to 3.5 >3.5 to 4.28 >4.28 to 5.3 >5.3
Zannat et al. (2019) None Perceived PLOS score Not defined as per service levels
TRANSPORT REVIEWS 13
5.2. Attributes most repeated
Attributes involved in 47 studies were counted for repetitions and classified as per their
association with the broad constructs of flow characteristics (FC), built environment (BE)
and users’perception (UP).The major challenge faced in calculating the frequency of
each attributes was in identifying the different names assigned to essentially the same
attribute in different studies. 389 different attributes are used in the 47 studies, which
were classified and grouped as 44 distinct attributes.
Figure 5 suggests that attribute such as “Flow rate of pedestrians”, which has been
classified as FC type attribute, have the highest repetition across the 47 studies. It is to
be noted, however, that studies emanating from India use both flow rate and pedestrian
space (area module) extensively. Among the BE attribute type, “width of sidewalk”is the
most widely used attribute, whereas among UP attribute type it is the “presence of lateral
separation”that is often repeated. However, specifically for studies originating in India,
“aesthetics”was the most used attribute under the UP broad construct category. The
results were compared to that of the Raad and Burke (2018) review, which validated
that “width of sidewalk”and “pedestrian flow rate”were indeed the most used utilised
attributes.
It is worth mentioning that attributes are classified under FC, BE and UP based on their
approach of data collection. FC and BE constructs are quantitative whereas UP is qualitat-
ive in nature. FC and BE attributes are actual measurements of the features from the
ground and UP attributes are views, documented from users of the facility. To clarify
further, “presence of sidewalk”,“presence of lateral separation”, etc., although may be a
BE feature, but was classified as UP, since the studies using these attributes recorded
them as response of users’satisfaction.
5.3. Broad constructs involved in the studies
As is shown in Table 2 (and Appendix 1), each study was classified as FC, BE, UP or a
combination thereof, based on the use of attributes pertaining to the constructs and
their method of data collection. If they involved quantitative attributes like FC or BE
solely, they were classified as FC or BE respectively; alternately if the studies involved
qualitative variables like UP in conjunction with a quantitative attribute FC then such
studies are classified under “UP + FC”. There were studies which seemed to have uti-
lised attributes from all three constructs, however they were not used in a combined
fashion to estimate the PLOS measure. An example of such classification seen in
Table 2 is “FC & BE + UP”. It can be seen from Figure 6, that the highest share of
studies were the ones that utilised attributes related to FC (13 out of 47) followed
by studies that utilised attributes pertaining to the “UP & FC + BE”(11 out of 47)
category.
It was also seen that 6 out of the 22 (i.e. 28%) studies attributed to developing
economies, were related to the “UP & FC + BE”category. The share of studies in this
category is the highest, all others being –FC (22%), BE (14%), UP (14%), “FC & BE +
UP”(9%), “FC + BE”(4%), and “BE + UP”(9%). Clearly, “UP & FC + BE”category is a
popular broad construct category amongst newly created PLOS models emanating
from developing countries.
14 D. NAG ET AL.
Figure 5. Most utilised (a) flow characteristics; (b) built environment; (c) users’perception attribute
type.
TRANSPORT REVIEWS 15
5.4. Analysis techniques
Referring to Figure 7, we see that out of the 47 research items, 28 of them used regression
to estimate their PLOS model. The linear regression technique was utilised mainly for esti-
mating the flow parameter relationships whereas multiple linear regression (MLR) was
used for quantifying pedestrian perception and analysing relationships. Clearly, regression
is the most preferred technique for estimation, especially MLR; however, the assumed
relationship may not be linear.
There are 10 studies which used an analytical points scale system (refer Figure 7). Asadi-
Shekari, Moeinaddini, and Zaly Shah (2014) and Christopoulou and Pitsiava-Latinopoulou
(2012) utilised guidelines from around the world to create weightages for the assessment
criteria in order to make their evaluation robust. Techniques listed under the “others”cat-
egory include –conjoint analysis, structural equation modelling, marginal effects models,
artificial neural networks, clustering algorithm, and data visualisation techniques. One of
the studies used alternate modelling techniques (Zhao, Bian, Rong, Liu, & Shu, 2016),
where they utilised a fuzzy neural network, showed not only a more accurate model for
PLOS estimation, but also helped in the scientific categorisation of service level.
PLOS studies from developing countries were found to be inclined towards multiple
linear regression (28%) and analytical points scale system (22%).
5.5. Service-level categorisation
PLOS estimated using statistical techniques and analytical points system yields final values
which are then categorised manually into LOS A, LOS B, LOS C, etc. But, manual interven-
tion in this process is defining the service levels incorrectly as shown by Sahani and
Bhuyan (2015). As seen from Table 3 (and Appendix 2), only 5 out of the 47 (Sahani,
2013; Sahani, Ojha, & Bhuyan, 2017; Sahani & Bhuyan, 2015,2017; Zhao et al., 2016)
Figure 6. Break-up of broad constructs involved in the 45 studies reviewed.
16 D. NAG ET AL.
studies have used advanced modelling techniques like fuzzy analysis, neural network and
clustering algorithm to categorise the pedestrian service levels. The use of such advanced
soft computing and statistical tool will not only yield more accurate results but also
address the quantification of cut-offvalues (threshold values) as posed by Kadali and
Vedagiri (2016). The developing countries considered this problem to be significant. All
the PLOS studies that have used a classification technique were seen to be emanating
from the developing nations.
6. Critical discussions and conclusions
A systematic review of 47 PLOS studies and 8 review papers was conducted with an aim to
highlight the similarities/differences among the studies thereby updating the readers with
the state-of-the-art practices on sidewalk PLOS. This review considers the PLOS develop-
ment process as consisting of three distinct steps –(a) attribute selection; (b) model devel-
opment; and (c) service-level categorisation. Limitations of previous PLOS models and the
way forward in the development of newer PLOS models are also discussed in the following
subsections.
6.1. Attribute selection for PLOS development
The review found 389 attributes used across 47 studies. These attributes were not only
duplicate in nature, but were also utilising inconsistent terminology. Moreover, Table 2
(and Appendix 1) suggests that there is a lack of a common methodology that would
Figure 7. Break-up (in terms of number of studies) of analysis techniques involved.
TRANSPORT REVIEWS 17
otherwise help researchers/practitioners in selecting suitable attributes for their PLOS
studies. 46 out of 47 studies have picked their attributes from existing literature
without any rigour to determine if those attributes are reflective of the study area’s
walking environment. Only Hidayat, Choocharukul, and Kishi (2011) have sought
users’opinion on 27 factors and checked their suitability using factor analysis. This
helps in understanding the relevance of the attribute as perceived by users. For
example, the factor analysis by Hidayat et al. (2011) helps the readers to realise that
“vendor attraction”and “vendor problems”on sidewalk facilities are pertinent factors
for determining performance of such facilities, and this is especially relevant for devel-
oping economies.
Users’opinion may be further augmented by collecting experts’views in identifying
relevant factors for PLOS studies. This is because the PLOS measures are developed for
the usage and interpretation by practitioners and professionals, which in-turn would
help in designing interventions specific to the study area. Hence both users and
experts are important actors in PLOS development and it would be
intuitive to examine their opinion early on in the process i.e. during the attribute selec-
tion stage.
Studies have portrayed robust methodologies and used advanced techniques for cali-
brating PLOS models, however, similar rigour has not been seen while selecting attributes
for developing the PLOS. To avoid inputting irrelevant attributes at the beginning of the
process, there is a need for taking a step backward and focus on attribute selection
comprehensively.
6.2. Under-representation of the three broad constructs in conjunction
PLOS studies have exhibited a wide variety of approaches between 1971 and 2019; begin-
ning by closely mimicking the methodology of the traditional vehicular LOS, to a com-
bined quantitative and qualitative estimation of PLOS. In some cases, researchers were
visually assessing the environment, whereas others were recording assessments based
on the users. Early PLOS measures (1970s and 1980s) were related to the broad constructs
of FC, as they were solely using pedestrian/trafficflow parameters (Fruin, 1971; Polus,
Schofer, & Ushpiz, 1983; Pushkarev & Zupan, 1975; Tanaboriboon & Guyano, 1989).
Later on (1980s onward till late 1990s) qualitative PLOS assessment of the built environ-
ment, which relates to BE, was carried out by researchers (Dixon, 1996; Gallin, 2001;
Jaskiewicz, 2000; Sarkar, 1994). This was the phase when researchers understood that
such visual assessments were subjective, and hence began to record users’perception
(UP) (Khisty, 1994).
All these approaches were state-of-the-art before 2000s. However, Landis’model
(Landis et al., 2001) seemed to be the turning point of PLOS studies. Landis et al. (2001)
advocated a transferrable PLOS, which was based on the roadside characteristics, and
developed a model through MLR. This study was the first which objectively quantified
pedestrian perception along a roadway segment using measurable traffic and roadway
variables. They utilised the mathematical expression shown in Equation (1) to carry out
stepwise regression. Subsequently researchers used built environment and traffic par-
ameters to characterise their PLOS model thereby taking a holistic approach by utilising
18 D. NAG ET AL.
more than just one broad construct.
PLOS Score =a1×f(lateral separation factors) +a2×f(traffic volume)
+a3×f(speed, vehicle type)
+a4×f(driveway access frequency and volume) +Constant
(1)
Following Landis et al.’s development, several other studies started developing and
experimenting with their approach of modelling, where users’perception was considered
as a dependent variable. Researchers applied this to different Asian country contexts as
well, such as China (Meng, Zhu, & Zeng, 2014; Tan, Wang, Lu, & Bian, 2007; Zhao et al.,
2016), India (Marisamynathan & Lakshmi, 2016; Sahani et al., 2017), Thailand (Hidayat
et al., 2011) and Malaysia (Daniel et al., 2016). While this approach of modelling PLOS is
very intuitive and practical, it must be noted that practitioners applying this technique
do not collect users’perception to arrive at the potential PLOS score. Looking at Equation
(1), the developed model requires only visual assessment inputs of built physical charac-
teristics and trafficflow parameters to predict users’perception, which is then classified
into PLOS levels. Thus one could argue that such studies partially utilise the broad con-
structs, i.e. combining BE and FC to predict UP, rather than a combination of BE, FC and
UP to predict PLOS.Reviewing the broad constructs involved in all studies as shown in
Table 2 (and Appendix 1), it is seen that there is a large variation in the use of attributes
that relates to the three broad construct. Out of 47 studies, 29% were found to be associ-
ated with flow characteristics (FC), 18% with built environment (BE), and 11% with users’
perception (UP). Remaining 42% studies were in combination with any two of the three
broad constructs. None of the studies were found to be associated with all three broad
constructs in conjunction. This suggests that a combined method needs to be developed
taking into account all the three broad constructs.
6.3. Modelling and classification approaches for developing PLOS
There is a consistency across studies regarding techniques used for estimating the PLOS
model. It was found that statistical regression was the most favoured, with 60% of the
studies using this technique. Within the regression-based studies, simple linear regression
(SLR) and multiple linear regression (MLR) methods are the most popular ones. The good-
ness of fit(R
2
) values for these models varied in the range of 0.21–0.972. Over-reliance of
these techniques in PLOS studies compels us to critique the linearity between dependent
and independent attributes used in such regressions, which may not always hold true. Fur-
thermore, the LOS concept has been regarded as a classification problem by researchers
(Bhuyan & Nayak, 2017), but for classifying service levels, only 11% of studies utilised a
classification technique based on MOEs. Remaining studies perform this categorisation
manually, using the judgment of the researcher, and thus the service-level definition
varies from study to study. For example, as seen from Table 3 (and Appendix 2),
Dowling et al. (2009), Landis et al. (2001) and Petritsch et al. (2006) had similar definition
of service levels i.e. letter-grade “B”is assigned when the score is in between 1.5 and 2.5,
“C”when the score is in between 2.5and 4.5 and so on; whereas Tan et al. (2007), who had
developed a PLOS measure similar to the Landis’method, used a different service-level
definition –“B”when the score is in between 2.0 and 2.5, “C”when the score is in
between 2.5 and 3.0 and so on. Here the model score interval for each service-level
TRANSPORT REVIEWS 19
definition is equal (i.e. the difference between upper limit and lower limit of model scores
for each service level remains constant). When Sahani et al. (2017) apply Genetic Program-
ming clustering algorithm to the outputs on similar PLOS models, service-level definitions
changes as per the result of the clustering algorithm and the model score interval does not
remain equal. This indicates that equal service-level definition does not hold true, and as
such studies that utilise a categorisation technique provides a more efficient service-level
categorisation rather than studies utilising researchers’judgement.
Machine learning models may be better techniques for developing PLOS measures.
Such models use the computational strength which has reached fruition in recent
years. Computers are made to “learn”and classify samples by “training”them accord-
ingly. There are two types of classification –supervised and unsupervised. Supervised
classification are done when the dependent variable can be observed and is used is
the modelling process. PLOS studies (Sahani & Bhuyan, 2017; Zhao et al., 2016) which
employ such classification tend to utilise users’perceived overall score (broad con-
struct is UP) that was recorded during the data collection process. These studies
used UP much like Landis’method. Unsupervised classification is employed when
the dependent variable is unobserved from the process. The output is generated
by understanding patterns within the dataset, this allows UP, BE and FC to be
included in the modelling process, as a part of the raw dataset being analysed.
Such an approach could be beneficial because –(a) the service levels are generally
unobserved and should be interpretable from the three constructs; and (b) it bypasses
the need to classify service levels manually, which was the case, seen from previous
studies.
6.4. Role of network attributes in walking environment evaluation
An important feature of the built environment that is currently not being captured by
existing PLOS studies in the assessment of pedestrian networks. However, there exist
studies that measure network connectivity and accessibility (Bandara, Wirasinghe, Gur-
ofsky, & Chan, 1994; Dill, 2004; Hillier, Penn, Grajewski, & Xu, 1993; Raford & Ragland,
2004; Tal & Handy, 2012) based on the configurational structure of links and nodes in
the pedestrian network. Hillier et al. (1993) have argued the importance of network
configuration and its relative effects on both pedestrian movement and landuse attraction.
According to them, “attraction theory”(i.e. theories on movement of people, to and from
different built forms or landuse with different degree of attractions) does not talk about
the spatial configuration of the urban links. There is ample evidence from existing litera-
ture that network plays an important role in route choice behaviour of pedestrians (Li &
Tsukaguchi, 2005; Muraleetharan & Hagiwara, 2007). As connectivity of pedestrian
network increases, travel distances decrease and route options increase, “allowing more
direct travel between destinations, and thereby, creating a more accessible and resilient
system”(Litman, 2004). Until now, the discussion on PLOS evaluation had been limited
to link level analyses, as was done for vehicular LOS assessment. An average pedestrian
trip-length ranges between 650 m and 1050 m (Arasan, Rengaraju, & Rao, 1994; Johar,
Jain, Garg, & Gundaliya, 2015; Rastogi & Krishna Rao, 2003) which is unlike vehicles that
can cover much longer distances. This aspect of a pedestrian trip is essential for a
network-wide assessment since there is a higher likelihood that links would be
20 D. NAG ET AL.
homogenous within short walking distances, rather than long driving links, and thereby
making the development of a network-wide PLOS much more meaningful.
Advances in Space Syntax techniques and analytic methods have been very impactful
in the past two decades. The ability of space-syntax based measure has been acknowl-
edged to be more impactful than environmental attributes of the built environment (i.e.
measurements and landuse characteristics) (Sharmin & Kamruzzaman, 2018). Therefore,
utilisation of such indices might be helpful for developing more efficient PLOS models.
This view was supported by Raad and Burke (2018) where the authors acknowledge
that connectivity is an under-researched aspects in PLOS studies and Space Syntax mod-
elling techniques may help achieve more meaningful results.
7. Contribution of this review study to the body of knowledge
This paper builds upon the existing 6 review studies and 47 research papers that have
been published worldwide on the topic of PLOS. There are three distinct aspects of this
paper that adds to the body of knowledge on PLOS, and by doing so, distinguishes
itself from the existing review studies. Firstly, this study updates readers with the most
recent advancements, specifically in the field of sidewalk PLOS measures, by reviewing
37 peer-reviewed journal articles, 6 indexed conference proceeding articles and 4 techni-
cal documents. The review range spanned from 1971 up to 2019, whereas the latest pub-
lished review article’s range was from 1971 to 2017. Findings from the six review articles
along with two other related review papers, on walkability indices, were also examined
and reported. Such a review of sidewalk PLOS studies in conjunction with walkability
indices has not been undertaken till date.
Secondly, this study reviews each PLOS study based on their distinct components. The
authors are of the view that each PLOS development follow three steps –(a) attribute
selection for PLOS; (b) modelling and calibration of the measure; and (c) service-level cat-
egorisation of the outputs. This component-based approach to understanding PLOS was
found to be absent in the other review papers even though the three components are
assessed in each of the 47 PLOS papers. By breaking it down in to three components,
the readers get an in-depth and clear understanding of the PLOS development process.
Table 2 (and Appendix 1) lists the PLOS studies as per their attribute selection procedure
and modelling approach, whereas, Table 3 (and Appendix 2) gives information about the
service-level definitions. Table 2 tells the readers how most studies do not follow any attri-
bute selection procedure, whereas Table 3 allows readers to compare service-level
definitions of different studies, which was non-existent in the reviews till date. Further-
more, Table 3 exhibits that most studies categorised their service levels manually, which
may not be giving consistent and robust results.
Finally, this review evaluates each PLOS study from the point of view of the three broad
constructs, and strongly advocates the usage of a PLOS measure which is associated with
all the constructs in conjunction. Table 2 (Appendix 1) categorised each PLOS study as per
their association with the broad constructs and it was interesting to note that none of the
PLOS studies were associated with the three constructs in a combined manner. Further-
more, this study also identifies a specific aspect of the built environment (one of the
three broad construct) that is not being discussed in most review papers. This aspect is
the pedestrian network and is shown to be an important component of the walking
TRANSPORT REVIEWS 21
experience that can hamper the performance of the built walking environment. Therefore,
the discussions included the possibility of defining PLOS measures based on network attri-
butes and expand the PLOS definition over a network rather than just links.
Disclosure statement
No potential conflict of interest was reported by the authors.
Funding
This paper is a part of a research project titled “Smart and Integrated Pedestrian Network Design”
under the “Uchhatar Avishkar Yojana (UAY)”scheme. The Ministry of Human Resource & Develop-
ment (MHRD), Govt. of India; Ministry of Urban Development (MoUD), Govt. of India; Vikram Solar
Pvt. Ltd. and GMR Airport Developers Ltd. have jointly funded the project.
ORCID
Dipanjan Nag http://orcid.org/0000-0002-1192-2161
Arkopal Kishore Goswami http://orcid.org/0000-0003-1369-215X
Ankit Gupta http://orcid.org/0000-0003-1789-9502
Joy Sen http://orcid.org/0000-0002-4605-9273
References
Arasan, V. T., Rengaraju, V. R., & Rao, K. V. K. (1994). Characteristics of trips by foot and bicycle modes
in Indian city. Journal of Transportation Engineering,120(2), 283–294.
Asadi-Shekari, Z., Moeinaddini, M., & Zaly Shah, M. (2012). Disabled pedestrian level of service
method for evaluating and promoting inclusive walking facilities on urban streets. Journal of
Transportation Engineering,139(February), 181–192. doi:10.1061/(ASCE)TE.1943-5436.0000492
Asadi-Shekari, Z., Moeinaddini, M., & Zaly Shah, M. (2013). Non-motorised level of service: Addressing
challenges in pedestrian and bicycle level of service. Transport Reviews,33(2), 166–194. doi:10.
1080/01441647.2013.775613
Asadi-Shekari, Z., Moeinaddini, M., & Zaly Shah, M. (2014). A pedestrian level of service method for
evaluating and promoting walking facilities on campus streets. Land Use Policy,38, 175–193.
doi:10.1016/j.landusepol.2013.11.007
Bandara, S., Wirasinghe, S. C., Gurofsky, D., & Chan, P. (1994). Grade-separated pedestrian circulation
systems. Transportation Research Record,1438,59–66.
Bhuyan, P. K., & Nayak, M. S. (2017). A review on level of service analysis of urban streets, 1647(April).
doi:10.1080/01441647.2013.779617
Bivina, G. R., & Parida, M. (2019). Modelling perceived pedestrian level of service of sidewalks: A struc-
tural equation approach. Transport,34(3), 339–350. doi:10.3846/transport.2019.9819
Bivina, G. R., Parida, P., Advani, M., & Parida, M. (2018). Pedestrian level of service model for evaluating
and improving sidewalks from various land uses. European Transport –Trasporti Europei,67,1–18.
Bloomberg, M. R., & Burden, A. M. (2006). New York City pedestrian level of service study phase I.
New York. Retrieved from https://www1.nyc.gov/assets/planning/download/pdf/plans/
transportation/td_fullpedlosb.pdf
Bunevska Talevska, J., & Malenkovska Todorova, M. (2012). Contribution to sidewalk pedestrian level
of service analysis. Transport Problems,7(3), 5–13.
Campos, B. (2012). Book review on Jan Gehl-Life between buildings: Using public space. The Journal
of Space Syntax,3(1), 125–128. Retrieved from http://books.google.hu/books/about/Life_
Between_Buildings.html?id=K98JAQAAMAAJ&pgis=1
22 D. NAG ET AL.
Central Road Research Institute. (2018). Indian highway capacity manual (Indo-HCM). New Delhi.
Cepolina, E. M., Menichini, F., & Gonzalez Rojas, P. (2018). Level of service of pedestrian facilities:
Modelling human comfort perception in the evaluation of pedestrian behaviour patterns.
Transportation Research Part F: Traffic Psychology and Behaviour,58, 365–381. doi:10.1016/j.trf.
2018.06.028
Cervero, R., Sarmiento, O. L., Jacoby, E., Gomez, L. F., & Neiman, A. (2009). Influences of built environ-
ments on walking and cycling: Lessons from Bogotá. International Journal of Sustainable
Transportation,3(4), 203–226. doi:10.1080/15568310802178314
Choi, J., Kim, S., Min, D., Lee, D., & Kim, S. (2016). Human-centered designs, characteristics of urban
streets, and pedestrian perceptions. Journal of Advanced Transportation,50, 120–137. doi:10.
1002/atr.1323
Christopoulou, P., & Pitsiava-Latinopoulou, M. (2012). Development of a model for the estimation of
pedestrian level of service in Greek urban areas. Procedia –Social and Behavioral Sciences,48,
1691–1701. doi:10.1016/j.sbspro.2012.06.1144
Daniel, B. D., Mohamad Nor, S. N., Md Rohani, M., Prasetijo, J., Aman, M. Y., & Ambak, K. (2016).
Pedestrian footpath level of service (FOOT-LOS) model for Johor Bahru. In The 3rd International
Conference on Civil and Environmental Engineering for Sustainability (Vol. 47, pp. 3006–3011).
doi:10.1051/matecconf/20164703006
Davis, D. G., & Braaksma, J. P. (1987). Level-of-service standards for platooning pedestrians in trans-
portation terminals. ITE Journal (Institute of Transportation Engineers),57(4), 31–35.
Dill, J. (2004). Measuring network connectivity for bicycling and walking. In 83rd Annual Meeting of
the Transportation …(p. 20). Washington, DC. Retrieved from http://reconnectingamerica.org/
assets/Uploads/TRB2004-001550.pdf
Dixon, L. (1996). Bicycle and pedestrian level-of-service performance measures and standards for
congestion management systems. Transportation Research Record: Journal of the Transportation
Research Board,1538,1–9. doi:10.3141/1538-01
Dowling, R., Flannery, A., Landis, B., Petritsch, T., Rouphail, N., & Ryus, P. (2009). Multimodal level of
service for urban streets. Transportation Research Record: Journal of the Transportation Research
Board,2071(2071), 1–7. doi:10.3141/2071-01
Ewing, R., & Cervero, R. (2001). Travel and the built environment: A synthesis. Transportation Research
Record: Journal of the Transportation Research Board,1780(01), 87–114. doi:10.3141/1780-10
Ewing, R., & Cervero, R. (2010). Travel and the built environment. Journal of the American Planning
Association,76(3), 265–294. doi:10.1080/01944361003766766
Ewing, R., Hajrasouliha, A., Neckerman, K. M., Purciel-Hill, M., & Greene, W. (2016). Streetscape features
related to pedestrian activity. Journal of Planning Education and Research,36(1), 5–15. doi:10.1177/
0739456X15591585
Franceschini, F., Cignetti, M., & Caldara, M. (1998). Comparing tools for service quality evaluation.
International Journal of Quality Science,3(4), 356–367. doi:10.1108/13598539810243658.
Fruin, J. J. (1971). Designing for pedestrians: A level-of-service concept. Highway Research Record,355,
1–15.
Gallin, N. (2001). Quantifying pedestrian friendliness guidelines for assessing pedestrian level of
service. Road & Transport Research,10(1), 47–55.
Gupta, A., & Pundir, N. (2015). Pedestrian flow characteristics studies: A review. Transport Reviews,
35(4), 445–465. doi:10.1080/01441647.2015.1017866
Hajrasouliha, A., & Yin, L. (2015). The impact of street network connectivity on pedestrian volume.
Urban Studies,52(13), 2483–2497. doi:10.1177/0042098014544763
Hidayat, N., Choocharukul, K., & Kishi, K. (2011). Pedestrian level of service model incorporating ped-
estrian perception for sidewalk with vendor Activities. Journal of the Eastern Asia Society for
Transportation Studies,9(October 2011), 1012–1023. doi:10.1111/nmo.12900
Hillier, B., Penn, A., Grajewski, T., & Xu, J. (1993). Natural movement –or, configuration and attraction
in urban pedestrian movement. Environment and Planning B: Planning and Design,20(1), 29–66.
doi:10.1068/b200029
Indian Road Congress. (2012). Guidelines for pedestrian facilities. New Delhi, India. Retrieved from
https://go.itdp.org/.../IRC2012%28Guidelines For Pedestrian Facilities
TRANSPORT REVIEWS 23
International Monetary Fund. (2018). World economic outlook: Challenges to steady growth.
Washington, DC: World Economic Outlook.
Jaskiewicz, F. (2000). Pedestrian level of service based on trip quality. Transportation Research Record,
E-C019,1–14.
Jensen, S. U. (2007). Pedestrian and Bicyclist level of service on roadway segments. Transportation
Research Record,2031,43–51. doi:10.3141/2031-06
Johar, A., Jain, S. S., Garg, P. K., & Gundaliya, P. J. (2015). A study for commuter walk distance from bus
stops to different destination along routes in Delhi. European Transport –Trasporti Europei,59(3),
1–15.
Kadali, B. R., & Vedagiri, P. (2016). Review of pedestrian level of service perspective in developing
countries. Transportation Research Record: Journal of the Transportation Research Board,2581,
37–47. doi:10.3141/2581-05
Karatas, P., & Tuydes-Yaman, H. (2018). Variability in sidewalk pedestrian level of service measures
and rating. Journal of Urban Planning and Development,144(4). 04018042-1–11. doi:10.1061/
(ASCE)UP.1943-5444.0000483
Khisty, C. J. (1994). Evaluation of pedestrian facilities: Beyond the level-of-service concept. Transport
Research Record,1438,45–50.
Kim, S., Choi, J., Kim, S., & Tay, R. (2014). Personal space, evasive movement and pedestrian level of
service. Journal of Advanced Transportation,48, 673–684. doi:10.1002/atr
Kim, K., Hallonquist, L., Settachai, N., & Yamashita, E. (2006). Walking in Waikiki, Hawaii: Measuring
pedestrian level of service in an urban Resort District. Transportation Research Record: Journal of
the Transportation Research Board,1982, 104–112. doi:10.3141/1982-15
Kim, T., Sohn, D. W., & Choo, S. (2017). An analysis of the relationship between pedestrian traffic
volumes and built environment around metro stations in Seoul. KSCE Journal of Civil
Engineering,21(4), 1443–1452. doi:10.1007/s12205-016-0915-5
Landis, B. W., Ottenberg, R. M., Mcleod, D. S., & Guttenplan, M. (2001). Modeling the roadside walking
environment: A pedestrian level of service. Transportation Reasearch Record,1773(01), 1–27. doi:10.
3141/1773-10
Li, Y., & Tsukaguchi, H. (2005). Relationship between network topology and pedestrian route choice
behaviour. Journal of the Eastern Asia Society for Transportation Studies,6(1), 241–248.
Litman, T. (2004). Integrating public health objectives in transport decision-making. American Journal
of Health Promotion, Inc,18(1), 103–108. doi:10.4278/0890-1171-18.1.103
Maghelal, P. K., & Capp, C. J. (2011). Walkability: A review of existing pedestrian indices. Journal of
the Urban and Regional Information Systems Association,23(2), 5–19. doi:10.1017/
CBO9781107415324.004
Marisamynathan, S., & Lakshmi, S. (2016). Method to determine pedestrian level of service for sidewalks
in Indian context. Transportation Letters,7867(January), 1–8. doi:10.1080/19427867.2016.1264668
Meng, J. Q., Zhu, Z. J., & Zeng, J. (2014). The research on the system of level of service of the city com-
mercial pedestrian street. Applied Mechanics and Materials,505–506, 656–665. doi:10.4028/www.
scientific.net/AMM.505-506.656
Mori, M., & Tsukaguchi, H. (1987). A new method for evaluation of level of service in pedestrian facili-
ties. Transportation Research Part A: General,21(3), 223–234. doi:10.1016/S0304-5013(06)72617-6
Muraleetharan, T., Adachi, T., Uchida, K., Hagiwara, T., & Kagaya, S. (2004). A study on evaluation of
pedestrian level of service along sidewalks and crosswalks using conjoint analysis. Journal of
Infrastructure Planning, Japan Society of Civil Engineers,21(3), 727–735.
Muraleetharan, T., & Hagiwara, T. (2007). Overall level of service of urban walking environment and its
influence on pedestrian route choice behavior: Analysis of pedestrian travel in Sapporo, Japan.
Transportation Research Record: Journal of the Transportation Research Board,2002,7–17. doi:10.
3141/2002-02
Olsen, J. R., Macdonald, L., & Ellaway, A. (2017). Changes over time in population level transport sat-
isfaction and mode of travel: A 13 year repeat cross-sectional study, UK. Journal of Transport and
Health,6(February), 366–378. doi:10.1016/j.jth.2017.03.012
Ozbil, A., Peponis, J., & Stone, B. (2011). Understanding the link between street connectivity, land use
and pedestrian flows. Urban Design International,16(2), 125–141. doi:10.1057/udi.2011.2
24 D. NAG ET AL.
Park, K., Ewing, R., Sabouri, S., & Larsen, J. (2019). Street life and the built environment in an auto-
oriented US region. Cities,88(November 2018), 243–251. doi:10.1016/j.cities.2018.11.005
Parvathi, M. S. (2018). The analysis of factors affecting the pedestrian level of service on footpaths in
Uppal X road at Hyderabad. International Journal of Scientific Research and Review,7(9), 738–754.
Petritsch, T. A., Landis, B. W., Mcleod, P. S., Huang, H. F., Challa, S., Skaggs, C. L., …Vattikuti, V. (2006).
Pedestrian level-of-service model for urban arterial facilities with sidewalks. Transportation
Research Record: Journal of the Transportation Research Board,1982,84–89. doi:10.3141/1982-12
Polus, A., Schofer, J., & Ushpiz, A. (1983). Pedestrian flow and level of service. Journal of Transportation
and Engineering,109(1), 46–56. doi:10.1061/(ASCE)0733-947X(1983)109:1(46)
Pushkarev, B. S., & Zupan, J. M. (1975). Urban space for pedestrians: A report for the regional plan associ-
ation. Cambridge, MA: MIT Press. Retrieved from http://www.amazon.com/dp/0262160633
Raad, N., & Burke, M. I. (2018). What Are the most important factors for pedestrian level-of-service
estimation? A systematic review of the literature. Transportation Research Record,doi:10.1177/
0361198118790623
Raford, N., & Ragland, D. (2004). Space syntax: Innovative pedestrian volume modeling tool for ped-
estrian safety. Transportation Research Record: Journal of the Transportation Research Board,1878,
66–74.
Raghuwanshi, A. K., & Tare, V. (2016). Assessment of pedestrian level of service for mixed lane.
Research Journal of Engineering and Technology,7(1), 11. doi:10.5958/2321-581X.2016.00003.9
Rastogi, R., Chandra, S., & Mohan, M. (2014). Development of level of service criteria for pedestrians.
Journal of the Indian Roads Congress,June 2015,60–69.
Rastogi, R., & Krishna Rao, K. V. (2003). Travel characteristics of commuters accessing transit: Case
study. Journal of Transportation Engineering,129(6), 684–694. doi:10.1061/(ASCE)0733-947X
(2003)129:6(684)
Sahani, R. (2013). Level of service criteria of urban walking environment in Indian context using Cluster
analysis. Rourkela: National Institute of Technology.
Sahani, R., & Bhuyan, P. K. (2013). Level of service criteria of off-street pedestrian facilities in Indian
context using affinity propagation clustering. In Procedia –Social and Behavioral Sciences (Vol.
104, pp. 718–727). Elsevier B.V. doi:10.1016/j.sbspro.2013.11.166
Sahani, R., & Bhuyan, P. K. (2015). Pedestrian level of service criteria for urban off-street facilities in
mid-sized cities using SOM in ANN. Indian Highways,5–15. doi:10.3846/16484142.2014.944210
Sahani, R., & Bhuyan, P. K. (2017). Pedestrian level of service criteria for urban off-street facilities in
mid-sized cities. Transport,32(2), 221–232. doi:10.3846/16484142.2014.944210
Sahani, R., Ojha, A., & Bhuyan, P. K. (2017). Service levels of sidewalks for pedestrians under mixed
traffic environment using genetic programming clustering. KSCE Journal of Civil Engineering,21
(7), 2879–2887. doi:10.1007/s12205-017-0042-y
Sahani, R., Praveena, D., & Bhuyan, P. K. (2016). Use of multinomial logit model in evaluation of service
levels of pedestrian facilities. In TPMDC 2016 (pp. 1–10). Mumbai.
Sarkar, S. (1994). Determination of service levels for pedestrians, with European examples.
Transportation Research Record,1405,35–42.
Sharmin, S., & Kamruzzaman, M. (2018). Meta-analysis of the relationships between space syntax
measures and pedestrian movement. Transport Reviews,34(4), 524–550. doi:10.1080/01441647.
2017.1365101
Shekari, Z. A., & Zaly Shah, M. (2011). Practical evaluation method for pedestrian level of service in
urban streets. In International Transport Research Conference. Penang, Malayasia.
Singh, K., & Jain, P. K. (2011). Methods of assessing pedestrian level of service. Journal of Engineering
Research and Studies,2(1), 116–124.
Sisiopiku, V., Byrd, J., & Chittoor, A. (2007). Application of level-of-service methods for evaluation of
operations at pedestrian facilities. Transportation Research Record,2002(1), 117–124. doi:10.3141/
2002-15
Stangl, P. (2012). The pedestrian route directness test: A new level-of-service model. Urban Design
International,17(3), 228–238. doi:10.1057/udi.2012.14
TRANSPORT REVIEWS 25
Tal, G., & Handy, S. (2012). Measuring nonmotorized accessibility and connectivity in a robust ped-
estrian network. Transportation Research Record: Journal of the Transportation Research Board,
2299,48–56. doi:10.3141/2299-06
Tan, D., Wang, W., Lu, J., & Bian, Y. (2007). Research on methods of assessing pedestrian level of
service for sidewalk. Journal of Transportation Systems Engineering and Information Technology,
7(5), 74–79. doi:10.1016/S1570-6672(07)60041-5
Tanaboriboon, Y., & Guyano, J. A. (1989). Level-of-service standards for pedestrian facilities in
Bangkok. A Case Study. Journal of the Institute of Transportation Engineers,59(11), 39–41. https://
doi.org/citeulike-article-id:7100921
Transportation Research Board. (2000). Highway capacity manual. Washington, DC: National Research
Council.
Transportation Research Board. (2010). Highway capacity manual. Washington, DC.
Vale, D. S., Saraiva, M., & Pereira, M. (2016). Active accessibility : A review of operational measures of
walking and cycling accessibility. Journal of Transport and Land Use,9(1), 209–235. doi:10.5198/jtlu.
2015.593
Victoria Transport Policy Institute. (2009). Walkability improvements: Strategies to make walking con-
venient, safe and pleasant. Retrieved from https://www.vtpi.org/tdm/tdm92.htm
Zainol, R., Ahmad, F., Nordin, N. A., & Aripin, A. W. M. (2014). Evaluation of users’satisfaction on ped-
estrian facilities using pair-wise comparison approach. In IOP Conference Series: Earth and
Environmental Science (Vol. 18, pp. 1–6). doi:10.1088/1755-1315/18/1/012175
Zannat, K. E., Raja, D. R., & Adnan, M. S. G. (2019). Pedestrian facilities and perceived Pedestrian Level
of Service (PLOS): A case study of Chittagong Metropolitan area, Bangladesh. Transportation in
Developing Economies,5(2), 1–16. doi:10.1007/s40890-019-0078-4
Zhao, L., Bian, Y., Rong, J., Liu, X., & Shu, S. (2016). Evaluation method for pedestrian level of service on
sidewalks based on fuzzy neural network model. Journal of Intelligent & Fuzzy Systems,30(5), 2905–
2913. doi:10.3233/IFS-151753
26 D. NAG ET AL.
Modelling specifications
Research items
Attribute selection
technique Dependent attribute Independent attributes Analysis method
No. of locations/sample
size/observations
Goodness of fit/
validation measure
Broad construct/s involved: flow characteristics
Fruin (1971) CEL Speed Area Module LR NR NR
Polus et al. (1983) CEL Model 1, 2, 3, 4:
Speed
Model 1, 2, 3, 4:
Density
LR 6 locations Model 1: R
2
= 0.602
Model 2: R
2
= 0.697
Model 3: R
2
= 0.918
Model 4: R
2
= 0.941
Davis and Braaksma
(1987)
CEL Time headway Speed Polynomial Regression 3 locations R
2
= 0.83
Tanaboriboon and
Guyano (1989)
CEL Speed Density LR NR NR
Highway Capacity
Manual (2000)
CEL Model 1: Speed;
Model 2: Flow rate;
Model 3: V/C ratio
Model 1, 2, 3, 4: Pedestrian space LR NR NR
Kim, Hallonquist,
Settachai, and
Yamashita (2006)
CEL NA Flow rate, pedestrian space,
sidewalk width, V/C ratio
Descriptive analysis and
data visualisation
16 screen lines NA
Bunevska Talevska
and Malenkovska
Todorova (2012)
CEL Pedestrian space Body ellipse, speed, acceleration,
position
Side friction micro-
simulation
3 residential streets NR
Sahani and Bhuyan
(2013)
CEL Flow rate, space, speed
and V/C ratio
Effective width of sidewalk Follows HCM (2000);
Affinity Propagation
(AP) clustering
algorithm
31 locations; 3764
pedestrians observed
NR
Kim et al. (2014) CEL Evasive movements Effective width, volume, landuse
characters
MLR 28 locations; 468 video
samples of 5 mins each
R
2
= 0.77 Validated with
perceived PLOS of 216
users
CEL Effective width of sidewalk NR
(Continued)
Appendices
Appendix 1. Modelling information for PLOS studies reviewed
TRANSPORT REVIEWS 27
Continued.
Modelling specifications
Research items
Attribute selection
technique Dependent attribute Independent attributes Analysis method
No. of locations/sample
size/observations
Goodness of fit/
validation measure
Sahani and Bhuyan
(2015)
Flow rate, space, speed
and V/C ratio
HCM (2000) methodology
and Self Organising
Maps (SOM) using
Analytical Neural
Network (ANN)
Silhouette, Davies-Bouldin,
Clinski-Harabasz and
Dunn Index
Raghuwanshi and
Tare (2016)
CEL Average pedestrian space V/C ratio of pedestrians, vehicles,
pedestrian crossing time, and
parking factors
MLR 9 street sections NR
Sahani and Bhuyan
(2017)
CEL Flow rate, space, speed
and V/C ratio
Effective width of sidewalk HCM (2000) methodology
and three clustering
algorithms –AP, SOM in
ANN and Genetic
Algorithm (GA)
3764 pedestrians
observed
Clusters validated using a
number of index –
Silhouette, Davies-
Bouldin, Clinski-Harabasz
and Dunn
Cepolina et al. (2018) CEL Perceived comfort Interpersonal distances Local density method,
Voronoi diagram
395 pedestrians Validated in comparison
with HCM (2000)
Broad Construct/s involved: Built environment
Sarkar (1994) Researchers’insight NA NA Descriptive service levels Two cities of Europe NA
Dixon (1996) CEL NA sidewalk, conflicts, amenities,
vehicular LOS, maintenance, multi-
modality
PBS NR NA
Jaskiewicz (2000) CEL NA enclosure, path complexity,
articulation, transparency, buffer,
shade trees, and physical
components
PBS 12 locations for
assessment
NA
Gallin (2001) CEL and stakeholder
consultation
NA Width, obstructions, volume,
security, connectivity,
environment, facilities, surface
quality, mix of users, conflict
PBS Not reported NA
Shekari and Zaly Shah
(2011)
Citing from 20
design guidelines
from developed
nations
NA Traffic speed, lanes, buffers, crossing
distance, and 13 other attributes
PBS Two collector urban
street
NA
Stangl (2012) CEL Route directness score Community block-size Pedestrian Route
Directness Index
8different block size
varying from 200 ft. X
200 ft. to 1000 ft. X
1000 ft.
NA
28 D. NAG ET AL.
Asadi-Shekari et al.
(2012)
CEL and design
guidelines
NA Model 1: Slope, elevator, curb ramp,
and 7 other attributes
Model 2: Traffic speed, pavement,
markings and 17 other attributes
PBS 1 street in Singapore NR
Asadi-Shekari et al.
(2014)
Citing from 20
design guidelines
from developed
nations
NA Traffic speed, lanes, buffers, mid
block and 23 other attributes
PBS 1 street in the campus –
20 guidelines reviewed
Not reported
Broad construct/s involved: users’perception
Khisty (1994) CEL NA Safety, security, comfort,
attractiveness, convenience,
coherence, continuity
PBS, constant-sum,
paired-comparison
method
302 responses NA
Muraleetharan,
Adachi, Uchida,
Hagiwara, and
Kagaya (2004)
CEL Users’rating of profiles Width & Separation, obstructions,
flow rate and bicycle events
Conjoint analysis 531 respondents Pearson’sR∼1 (near
perfect) and
Kendall’s Tau = 0.986
Muraleetharan and
Hagiwara (2007)
CEL Model 1: Respondents
rating
Model 2: Alternative
routes
Model 1: Width & Separation,
obstructions, flow rate and bicycle
events
Model 2: walking distance and LOS
score
Conjoint analysis &
Multinomial logit (MNL)
model
346 respondents and 215
sidewalks
Model 1: Not reported
Model 2: Pseodo-R
2
=
0.183, % of correct
prediction = 74%
Sahani et al. (2016) CEL Overall satisfaction score Traffic, safety, comfort, maintenance
and aesthetics score
MNL 1425 respondents Chi-square value =
505.5depicts good fit.
70% of data used for
model fit, remaining 30%
used to validate
Bivina et al. (2018) CEL NA Physical: Surface quality, width,
obstruction, vehicular conflict,
continuity, encroachment,
crossing facility, security, walk
environment, comfort
PBS 2804 respondents from 5
cities
Cronbach’s alpha > 0.7
assess internal
consistency
Bivina and Parida
(2019)
CEL Latent Exogenous: Safety;
Security; Mobility &
infrastructure; Comfort
& convenience
Latent Endogenous:
Perceived PLOS scores
from users
Traffic volume, traffic speed, police
patrolling, street lights, CCTV
camera, width of sidewalks,
continuity, encroachment, surface
quality, pedestrian amenities, bus
shelter, cleanliness, planning for
disabled, obstruction
Structural Equation
Modelling
502 responses Normed Fit Index (NFI) =
0.92; Comparative Fit
Index (CFI) = 0.953;
Tucker Lewis Index (TLI)
= 0.939; Acceptable is
NFI, CFI, TLI > 0.9
Root Mean Square Error
(RMSEA) = 0.05;
(Continued)
TRANSPORT REVIEWS 29
Continued.
Modelling specifications
Research items
Attribute selection
technique Dependent attribute Independent attributes Analysis method
No. of locations/sample
size/observations
Goodness of fit/
validation measure
Acceptable is RMSEA <
0.06
Broad constructs involved: users’perception + built environment
Christopoulou and
Pitsiava-
Latinopoulou (2012)
CEL NA Traffic factors, geometric/
environmental factors and
pedestrian movement factors
PBS Application on one
location
Compared with 5 existing
methodology
qualitatively
Parvathi (2018) CEL Perceived user score Sidewalk condition, road
characteristics, interaction
between pedestrians and other
modes, buffer, transit area and
safety
PBS, inverse variance to
calculate weights
Over 100 respondents NA
Zannat et al. (2019) CEL Model 1: Perceived PLOS
scores from users
Model 2: Perceived
PLOS scores from users
Model 3: Perceived
roadway conditions
Model 1: Perceived roadway
conditions –accessibility, safety,
comfort and attractiveness
Model 2: Physical feature
measurement –Footpath (width
of sidewalk, lighting, etc.)
carriageway (median width, guard
rail etc.) and transit facilities (bus
bay, sign, etc.)
Model 3: Physical feature
measurement
Model 1: Ordered probit
Model 2: Marginal
effects
Model 3: Multiple linear
regression
Model 1: 413 responses
Model 2: 413 responses
and physical feature
survey from 88 points
in the city
Model 2: 413 responses
and physical feature
survey from 88 points
in the city
NR for any model
Broad constructs involved: flow characteristics + built environment
Rastogi et al. (2014) CEL Model 1 & 2: LOS values
obtained from a
different study
Model 1 & 2: Pedestrian flow,
Sidewalk width, obstruction
Model 1 & 2: Stepwise
MLR and Pearson’s
correlation analyses
Sample size: 517 Model 1: R
2
= 0.899
Model 2: R
2
= 0.899
Karatas and Tuydes-
Yaman (2018)
CEL Pedestrian volume,
assessment score
Density, walkway width, buffer area,
shade, enclosure, motor vehicle,
maintenance, conflicts
Minimum PLOS of the
three methodology or
weighted average
method
81 road segments NA
Broad construct/s involved: (a) flow characteristics (b) users’perception + built environment
Mori and Tsukaguchi
(1987)
CEL Model 1: speed
Model 2: Participants
scores
Model 1: density
Model 2: effective sidewalk width,
green ratio, sidewalk type
Model 1: LR
Model 2: MLR
Model 1: NR
Model 2: 35
respondents
Model 1: NR
Model 2: R
2
= 0.85
CEL NR NR
30 D. NAG ET AL.
Highway Capacity
Manual (2010)
Model 1: Participants
scores
Model 2: Pedestrian
spacing
Model 1: Effective width of sidewalk,
width of bicycle lane, buffer,
vehicular traffic speed and
volume, presence of curb;
Model 2: Effective width of
sidewalk, width of bicycle lane,
lateral
Separation, walking speed
Model 1:
Step-wise MLR
Model 2: LR
Indian Road Congress
(2012)
CEL NA Pedestrian space, flow rate, speed.
Qualitative descriptions of built
environment
Quantitative –PLOS
values adopted from
HCM (2000)
Qualitative –description
of LOS values
NR NR
Indian Highway
Capacity Manual
(2018)
CEL Model 1: flow rate
Model 2: Speed
Model 3: Space
Model 4: Users’score
Model 1, 2, 3: Density
Model 4: QoS (Bivina et al., 2018):
Physical and users’characteristics
Model 1, 2, 3: LR
Model 4: Bivina et al.,
(2018)–PBS, weighted
average method
Model 1, 2, 3: sample size
= 951
Model 4: NR
NR
Broad construct/s involved: (a) users’perception (b) built environment + flow characteristics
Landis et al. (2001) CEL Respondents score Lateral separation factors, motor
vehicle volume, motor vehicle
speed, motor vehicle mix,
driveway access frequency and
volume
Stepwise MLR 75 respondents; 42
segments
R
2
= 0.85
Petritsch et al. (2006) CEL Respondents score Crossing width (per mile) at conflict
locations, Average 15-min volume
of vehicular traffic adjacent to the
sidewalk
Pearson’s correlation
analysis, hypothesis
tests, and MLR
506 participants R
2
= 0.70
Jensen (2007) NR Share of Participants
stating a particular
score
Vehicular traffic volume, speed, type
of facility, width of pedestrian
facility, volume of pedestrians,
presence of trees, parked cars and
medians
Cumulative Logit Model
stepwise regression
407 participants R
2
= 0.55
Max. rescaled R
2
=0.57
Tan et al. (2007) CEL Response of users Pedestrian and bicycle volume,
driveway access frequency,
distance between sidewalk and
vehicle lane
Step-wise MLR 395 observations and 12
roadway segment
NR
Dowling et al. (2009) NR Response of users Effective width of sidewalk, buffer,
speed of vehicles, volume of
vehicles, curb
MLR 140 respondents were
showed 90 video clips
NR
(Continued)
TRANSPORT REVIEWS 31
Continued.
Modelling specifications
Research items
Attribute selection
technique Dependent attribute Independent attributes Analysis method
No. of locations/sample
size/observations
Goodness of fit/
validation measure
Hidayat et al. (2011) Factor analysis of 27
factors from
existing literature
based on users’
opinion
Model 1, 2, 3:
Perception of
pedestrian
Model 1: comfort, vendor attraction
Model 2: comfort, vendor
problems
Model 3: comfort, vendor
problems, pedestrian volume,
interaction vendors
MLR 1072 respondents from
Bangkok; 523 from
Jakarta
Model 1: R
2
= 0.27
Model 2: R
2
= 0.21
Model 3: R
2
= 0.24
Meng et al. (2014) CEL Model 1: Respondents
score (main)
Model 2: Cross-section
var. (comp. of 1)
Model 3:
Comprehensive var. of
street (comp. of 1)
Model 4: Log value of
Auxiliary street var.
(comp. of 1)
Model 1: Cross-section, ped. flow,
obstacles, comprehensive,
auxiliary
Model 2: width, elevation diff., and
separators exists
Model 3: quality of pavement
Model 4: utility variables like light,
shade, garbage bins, recreational
and public health
MLR 227 respondents from 2
locations
Model 1: R
2
= 0.4
Model 2: R
2
= 0.4
Model 3: R
2
= 0.4
Model 4: NR
Marisamynathan and
Lakshmi (2016)
CEL Overall satisfaction level Sidewalk surface, presence of
guardrails and barriers, traffic
volume, sidewalk width
Step-wise MLR 540 respondents R
2
= 0.935; 90% data used
for model fit 10% for
validation. MAPE = 3.14%
and RMSE = 2.02%
Zhao et al. (2016) CEL Users’perception scores Flow rate, vehicular flow, effective
width, segregation facilities,
frequency of barriers, on-street
parking, green looking ratio,
connected regions
Image processing and
Fuzzy neural network
87 sidewalks and 4300
responses
Accuracy = 0.94 compared
with the testing data set
Daniel et al. (2016) CEL Participants score Footpath width, road width, surface
damage, number of obstructions,
pedestrian flow traffic volume
MLR 391 respondents from 25
roads
R
2
= 0.97, Validation –
average error 1.28%
Sahani et al. (2017) CEL Model 1: Overall
Satisfaction
Model 2: Perception
scores
Model 1: Platoon size, traffic, safety,
comfort, maintenance and
aesthetic score
Model 2: width, pedestrian
volume, obstruction, motorised
and non-motorised volume
Factor analysis and
discriminant analysis for
selecting variables; step-
wise MLR for PLOS
Model; GP for LOS
classification
1825 respondents Model 1: NR
Model 2: R
2
= 0.972
Note: CEL: Citing from Existing Literature; LR: Linear Regression; MLR: Multiple Linear Regression; NA: Not Applicable; NR: Not Reported; PBS: Points-Based System
32 D. NAG ET AL.
Appendix 2. PLOS service-level definitions and classification techniques from different studies
Research items
Classification
technique used
MOE used for PLOS
classification
Service-level definitions
AB CD E F
Broad Construct/s involved: flow characteristics
Fruin (1971) None Area module (ft
2
/ped) >35 25–35 15–25 10–15 5–10 <5
Volume (ped/min/ft) 7 7–10 10–15 15–20 20–25 >25
Polus et al. (1983) None Density (ped/m
2
) <0.60 0.61–0.75 C
1
0.75–
1.25
C
2
1.26–2.00
Not defined Not defined Not defined
Area module (m
2
/ped) >1.67 1.66–1.33 C
1
1.33–
0.80
C
2
0.80–0.50
Not defined Not defined Not defined
Average speed (m/
min)
0–40 40–50 C
1
50–75
C
2
75–95
Not defined Not defined Not defined
Davis and Braaksma
(1987)
None Volume (ped/min/m) A+
<37
A
37–46
46–57 57–68 68–75 75–57 <57
Area module (m
2
/ped) A+
>2.3
A
1.7–2.3
1.3–1.7 1.0–1.3 0.8–1.0 0.7–0.8 <0.7
Speed (m/s) A+
>1.4
A
1.3–1.4
1.2–1.3 1.1–1.2 1.0–1.1 0.7–1.0 <0.7
Tanaboriboon and
Guyano (1989)
None Area module (m
2
/ped) >2.38 1.60–2.38 0.98–1.60 0.65–0.98 0.37–0.65 <0.37
Volume (ped/min/m) <28 28–40 40–61 61–81 81–101 >101
Highway Capacity
Manual (2000)
None Space (m
2
/ped) >5.6 >3.7–5.6 >2.2–3.7 >1.4–2.2 >0.75–1.4 ≤0.75
flow rate (ped./m/m
2
)≤16 >16–23 >23–33 >33–49 >49–75 >75
speed (m/s) >1.30 >1.27–1.30 >1.22–1.27 >1.14–1.22 >0.75–1.14 ≤0.75
V/C ratio ≤0.21 >0.21–0.31 >0.31–0.44 >0.44–0.65 >0.65–1.0 >1.0
Kim et al. (2006) None Space (ft
2
/ped) >60 >40–60 >24–40 >15–24 >8–15 ≤8
Flow rate (ped/min/ft) ≤5>5–7>7–10 >10–15 >15–23 >23
Bunevska Talevska and
Malenkovska
Todorova (2012)
None Space (ft
2
/ped) >60 >40–60 >24–40 >15–24 >8–15 ≤8.0
Sahani and Bhuyan
(2013)
Affinity propagation
clustering
Volume (ped/s/m) ≤0.052 >0.052–0.065 >0.065–0.081 >0.081–0.095 >0.095–0.114 >0.114
Space (m
2
/ped) >17.78–14.42 >11.3–14.42 >8.24–11.3 >7.82–8.24 >5.3–7.82 ≤5.3
speed (m/s) >1.53 >1.36–1.53 >1.14–1.36 >0.93–1.14 >0.71–0.93 ≤0.71
(Continued)
TRANSPORT REVIEWS 33
Continued.
Research items
Classification
technique used
MOE used for PLOS
classification
Service-level definitions
AB CD E F
V/C ratio ≤0.4 >0.4–0.57 >0.57–0.76 >0.76–0.9 >0.9–1.0 >1
Kim et al. (2014) None Evasive movements
(ped/min/m)
≤3.58 ≤6.32 ≤10.13 ≤13.06 ≤19.00 Over 19.00
space (m
2
/ped) ≥3.30 ≥2.00 ≥1.40 ≥0.90 ≥0.38 <0.38
Flow rate (ped/min/m) ≤20 ≤32 ≤46 ≤70 ≤106 >106
speed (m/s) ≥1.25 ≥1.20 ≥1.15 ≥1.03 ≥0.67 <0.67
Sahani and Bhuyan
(2015)
Self-organising maps in
analytical neural
networks
Space (m
2
/ped) >15.67 >11.94–15.67 >9.07–11.94 >6.49–9.07 >4.48–6.94 ≤4.48
Flow rate (ped/s/m) ≤0.063 >0.063–0.081 >0.081–0.103 >0.103–0.133 >0.133–0.145 >0.145
Raghuwanshi and Tare
(2016)
None Space (m
2
/ped) >4.9 3.3–4.9 1.9–3.3 1.3–1.9 0.6–1.3 <0.6
Sahani and Bhuyan
(2017)
AP, GA-Fuzzy & SOM in
ANN Clustering
methods
Volume (ped/s/m) ≤0.061 >0.61–0.081 >0.081–0.104 >0.104–0.127 >0.127–0.146 >0.146
Space (m
2
/ped) ≥16.53 <16.53 to 13.06 <13.06–9.91 <9.91–7.25 <7.25–4.48 ≤4.48
Speed (m/s) >1.21 >1.03–1.21 >0.88–1.03 >0.78–0.88 >0.62–0.78 ≤0.62
V/C ratio ≤0.34 >0.34–0.52 >0.52–0.67 .0.67–0.84 >0.84–1.0 >1.0
Cepolina et al. (2018) None Density (ped./m
2
) No clear
definition,
density has
been
compared
with HCM
(2000) LOS
Broad construct/s involved: built environment
Sarkar (1994) None Qualitative
descriptions of LOS
Highest here Safety
Security
Comfort
Continuity
System
coherence
Attractiveness
Lowest here
Dixon (1996) None Scores assigned by
visual assessment of
facilities
17–21 14–17 14–11 11–73–7<3
Jaskiewicz (2000) None Final scores from visual
assessment
4.0–5.0 3.4–3.9 2.8–3.3 2.2–2.7 1.6–2.1 1.0–1.5
Gallin (2001) None Final scores from visual
assessment
>132 101–131 69–100 37–68 <36 Not defined
34 D. NAG ET AL.
Shekari and Zaly Shah
(2011)
None Weighted average
PLOS% score
80–100 60–79 40–59 20–39 1–19 0
Stangl (2012) None Pedestrian route
directness score
85–100% 45–84% 30–44% 23–29% 7–22% 0–6%
Asadi-Shekari et al.
(2012)
None PLOS score generated
from the Disabled
PLOS and General
PLOS
80–100 60–79 40–59 20–39 1–19 0
Asadi-Shekari et al.
(2014)
None Weighted average
PLOS% score
80–100 60–79 40–59 20–39 1–19 0
Broad construct/s involved: users’perception
Khisty (1994) None Users’perception of
satisfaction
5 points 4 points 3 points 2 points 1 points 0 points
Muraleetharan et al.
(2004)
None Utility scores Not defined as
per service
levels –utility
value
decreases
from A to F
(linear
relationship)
Muraleetharan and
Hagiwara (2007)
None Utility scores High Medium Low
Sahani et al. (2016)None Overall satisfaction
score
<1.5 1.5 to <2.5 2.5 to <3.5 3.5 to <4.5 4.5–5.5 >5.5
Bivina et al. (2018) None Model scores >125 100–125 75–99 50–74 25–49 <25
Bivina and Parida
(2019)
None Perceived PLOS score Not defined as
per service
levels
Broad constructs involved: users’perception + built environment
Christopoulou and
Pitsiava-Latinopoulou
(2012)
None Assessment score 42–35 <35–28 <28–21 <21–14 <14–7<7–0
Parvathi (2018) None Perception score 4.34502–
5.64542
5.64543–6.49512 6.49513–7.79349 7.79350–9.09538 9.09539–10.39522 10.39523–
11.8697
Zannat et al. (2019) None Perceived PLOS score Not defined as
per service
levels
(Continued)
TRANSPORT REVIEWS 35
Continued.
Research items
Classification
technique used
MOE used for PLOS
classification
Service-level definitions
AB CD E F
Broad constructs involved: flow characteristics + built environment
Rastogi et al. (2014) None Space (m
2
/ped) >5.00 >2.22–5.00 >1.43–2.22 >1.00–1.43 >0.69–1.00 <0.69
Flow rate (ped/min/m) ≤18 >18–35 >35–51 >51–66 >66–73 >73
Karatas and Tuydes-
Yaman (2018)
None Volume (ped./15 min) Conceptual
model
proposed
hence no
service-level
definition
provided
Scores from visual
assessment
Broad construct/s involved: (a) flow characteristics (b) users’perception + built environment
Mori and Tsukaguchi
(1987)
None Volume (ped/min/m) <20 20–78 78–108 >108 Not defined Not defined
Density (ped/m
2
) <0.2 0.2–0.8 0.8–1.5 >1.5 Not defined Not defined
Overall evaluation Not defined as
per service
levels
Highway Capacity
Manual (2010)
None, worse of both
PLOS grades is used as
final grade
Space (ft
2
/ped) >60 >40–60 >24–40 >15–24 >8–15 ≤8.0
participants’(model)
score
≤2.0 >2–2.75 >2.75–3.50 >3.50–4.25 >4.25–5.00 >5.00
Indian Road Congress
(2012)
None Volume (ped/min/m) ≤12 12–15 15–21 21–27 27–45 >45
Space (m
2
/ped.) >4.9 3.3–4.9 1.9–3.3 1.3–1.9 0.6–1.3 <0.6
Qualitative description
–users & built
Ideal walking
condition and
factors
affecting PLOS
minimal
Reasonable condition
exists, factors
affecting safety and
comfort exists
Basic condition
but significant
factors affecting
safety and
comfort also
exists
Poor condition,
safety and
comfort minimal
Walk condition
unsuitable
Severely
restricted
walking
environment
Indian Highway
Capacity Manual
(2018)
None (classified as per
landuse –only
commercial is shown
here)
Flow rate
(ped/min/m)
≤13 >13–19 >19–30 >30–47 >47–69 >69
QoS: Model score ≥124 <124–106 <106–70 <70–52 <52 Not defined
Broad construct/s involved: (a) users’perception (b) built environment + flow characteristics
Landis et al. (2001) None Respondents score ≤1.5 >1.5–2.5 >2.5–3.5 >3.5–4.5 >4.5–5.5 >5.5
36 D. NAG ET AL.
Petritsch et al. (2006) None Respondents score ≤1.5 >1.5–2.5 >2.5–3.5 >3.5–4.5 >4.5–5.5 >5.5
Jensen (2007) None Share of participants
with level of
satisfaction
>50% very
satisfied
>50% moderately
satisfied and <50%
very satisfied
>50% little
satisfied and <
50% moderately
satisfied
>50% little
dissatisfied and
<50% little
satisfied
>50% moderately
dissatisfied and
<50% little
dissatisfied
>50% very
dissatisfied
Tan et al. (2007) None Participants’scores <2.0 2.0 to <2.5 2.5 to <3.0 3.0 to <3.5 3.5 to <4.0 4.0 and above
Dowling et al. (2009) None Participants’scores <1.5 1.5 to <2.5 2.5 to <3.5 3.5 to <4.5 4.5 to <5.5 5.5 and above
Hidayat et al. (2011) None Participants’scores >9.0 >7.0–9.0 >5.0–7.0 >3.0–5.0 >2.0–3.0 2.0 or below
Meng et al. (2014) None Respondents’score ≤1.5 >1.5–2.5 >2.5–3.5 >3.5–4.5 >4.5 Not defined
Marisamynathan and
Lakshmi (2016)
None Level of satisfaction <15% 15–30% 30–45% 45–60% 60–85% >85%
Zhao et al. (2016) Fuzzy Neural Network Pedestrian satisfaction
scores
10–98 7 6 5 4–1
Daniel et al. (2016) None Respondents’
perception score
>8.5 >7.0–8.5 >6.0–7.0 >5.0–6.0 >4.0–5.0 ≤4.0
Sahani et al. (2017) Genetic Programming
(GP) clustering
Respondents’
perception score
≤1.8 >1.8–2.7 >2.7–3.5 >3.5–4.28 >4.28–5.3 >5.3
TRANSPORT REVIEWS 37